Pulse Modulation Techniques Questions and Answers

Answer: A

Because the average value of a width-modulated train is proportional to the message, a low-pass filter finishes the job, whereas position pulses are all identical and their average carries nothing. The bandwidths are comparable, synchronisation is still needed to define the slots, and position pulses amplify perfectly well. Converting into the form whose demodulator is trivial, rather than building a harder demodulator, is a recurring move in pulse circuit design.

Answer: B

The capacitor stores the captured voltage so that the output stays flat until the next sample, which is what makes the output flat-top rather than natural. Anti-alias filtering happens before the circuit, coupling is a separate function, and the aperture width is set by the switch's drive rather than by the capacitor. Its leakage and the following stage's input current between them decide how much the held value droops, which is the specification that matters in practice.

Answer: C

Droop means the held value is not quite the sampled value by the time it is used, which appears as an amplitude error and hence distortion. It does not change the sampling rate, cannot cause aliasing, which is settled before the sampler, and would reduce rather than increase the level presented to a quantiser. Any circuit asked to remember a value has to be specified for how well it remembers it, and here that specification is volts per second. Working one numerical case through both schemes side by side fixes the distinction better than any rule of thumb.

Answer: D

Aperture time is the window during which the switch is closed, so the signal's movement within it sets an uncertainty in the captured value - which is why a fast-moving signal needs a short aperture. Hold time, channel count and filter design are all separate parameters. The requirement scales with the signal's slope rather than its amplitude, which is the same quantity that governs slope overload in a delta modulator. Most errors on this topic come from carrying a result across from a neighbouring scheme where it does not hold.

Answer: A

Crosstalk is signal appearing where it should not, so the switch's isolation in the open state is what limits it. On-resistance affects amplitude accuracy and settling rather than isolation, current rating is a power consideration, and packaging is mechanical. Whenever a specification is about unwanted transfer rather than wanted transfer, the parameter to look up is the one describing the device's off state.

Answer: B

On a long route the decisive property is that each regenerator rebuilds the pulses cleanly, so noise does not accumulate hop by hop - an analogue pulse train degrades steadily with every amplifier. Digital coding needs more bandwidth rather than less, its coders were initially dearer, and it demands far more synchronisation. A scheme can win decisively on one property while losing on three others, which is why the application has to be named before any comparison is meaningful.

Answer: C

Power conversion and motor drives use it constantly, because a duty cycle controls average power directly and a switching device is either fully on or fully off, which is where its efficiency comes from. Long-distance voice and broadcast audio went digital. And deep space telemetry uses coded digital schemes with heavy error correction. A modulation scheme displaced from communication can remain the right answer in an entirely different field, and width modulation is the clearest example.

Answer: D

Time division lets one transmitter and one path serve dozens of sensors, each occupying its own slot, which is decisive when weight and power are limited. Identifying channels by amplitude would consume the parameter carrying the data. Synchronisation is required rather than dispensed with. And every channel needs anti-alias filtering whatever its bandwidth, since interference can be faster than the sensor. Sharing one expensive resource among many cheap sources is the economic argument behind all multiplexing.

Answer: A

Nyquist would allow 6.8 kHz, but a filter cannot fall from passband to stopband instantly, so the extra 1.2 kHz gives its transition band somewhere to live. It is not exactly twice the limit, the frame rate follows from the sampling rate rather than setting it, and quantisation noise depends on word length rather than sampling rate. Standard values almost always sit a little clear of a theoretical bound, and the margin is usually there for a component that cannot be ideal.

Answer: B

More levels means finer steps and so less quantisation noise, but each sample needs more bits and the rate rises accordingly - six decibels bought for one bit per sample. There is no way to have the improvement without the rate increase in a uniform coder, which is exactly why companding exists: it redistributes a fixed number of bits rather than adding more. Every improvement in a digital coder is paid for in bit rate, and the only question is what the bits buy.